Air-dielectric patch sector antenna for 868MHz
A single probe-fed patch made from sheet metal with air as the dielectric, for the EU 868-870 MHz band (LoRa, Meshtastic, Wireless M-Bus and similar). Simulated in CST Studio Suite 2016, built from zinc-coated steel sheet, measured with an R&S FSH6.
| Key data | Simulated | Measured |
|---|---|---|
| Match at 869 MHz (with radome) | S11 = -29 dB | S11 = -28.1 dB |
| Impedance at 869 MHz (with radome) | 54.3 - j0.9 Ω | 56.0 - j0.55 Ω |
| -10 dB bandwidth (with radome) | 42.2 MHz | 52 MHz |
| Gain | 9.61 dBi | not measured |
| 3 dB beamwidth, H-plane / E-plane | 67.9° / 54° | not measured |
| Front-to-back | ~22 dB | not measured |


Part 1: Simulation #
1.1 Concept #
- Radiator: rectangular patch, 12 mm above a ground plane, air dielectric (no dielectric loss, wide bandwidth, easy to cut by hand).
- Feed: coaxial probe (RG-401/U semi-rigid) from below, passing through the ground plane. The PTFE of the cable runs up to the patch, the inner conductor passes through the patch and is soldered on top.
- Polarization: linear, along the 151.5 mm patch axis (vertical when mounted).
- Cover: PVC radome, spaced 15 mm above the patch.
1.2 Final dimensions #
Coordinates: ground plane top at z = 0, origin at the patch center. The y axis is the resonant axis (feed axis, E-field).
| Parameter | Value | Meaning |
|---|---|---|
pl |
151.5 mm | patch length along y (resonant) |
pw |
190 mm | patch width along x |
pt |
0.6 mm | patch thickness |
h |
12 mm | gap between ground plane top and patch underside |
fy |
47 mm | probe offset from patch center along y (28.75 mm from the nearer edge) |
gw |
250 mm | ground plane width along x |
gl |
350 mm | ground plane length along y |
gt |
0.6 mm | ground plane thickness |
a |
0.815 mm | RG-401 inner conductor radius (1.63 mm diameter) |
b |
2.67 mm | RG-401 PTFE radius (5.34 mm diameter) |
| tube | 3.175 mm | RG-401 outer radius (6.35 mm diameter) |
cl |
30 mm or more | cable stub below the ground plane in the model |
ps |
1 mm | pin stick-out above the patch (real pin, soldered) |
| solder fillet | r = 2.5 mm, 1 mm high | model of the solder joint on top of the patch |
| radome | PVC, 15 mm above the patch | wall: 3 mm in the first model, 2 mm in the real build |
Check the RG-401 dimensions against your cable's datasheet. The coax impedance in the model is Z = 60/√εr · ln(b/a) with εr(PTFE) = 2.1, which gives about 50 Ω.



1.3 Building the model in CST #
- Define the parameters from the table above.
- Create the ground plane as a PEC brick (
gwxglxgt). - Create the PTFE cylinder (radius
b, εr = 2.1) from -clto 0 at (x = 0, y =fy) and cut the ground plane with it. - Add the outer tube (inner radius
b, outer radius 3.175 mm) from -clto -gt. - Create the patch brick (
pwxplxpt) at z =h. - Create the inner conductor (radius
a) from -clup through the patch toh+pt+ps, insert it into the PTFE and join it to the patch. - Add the solder fillet (cylinder, r = 2.5 mm, 1 mm high) on top of the patch around the pin.
- Add the PVC radome: PVC material, εr = 3.0, tan δ = 0.015 (typical values, not measured), open box on the ground plane, 15 mm above the patch.
- Define a waveguide port on the cable end (z = -
cl), covering the PTFE circle. - Frequency range 820-920 MHz, time domain solver, far-field monitor at 869 MHz.
Boundary conditions #
Set all six sides to "open (add space)", including the one below the ground plane.
Pitfall: an electric (PEC) boundary below the model sits only a few mm under the antenna and acts like an infinite ground plane. The results then look excellent but are wrong: no radiation behind the antenna (the gain at 180° shows -200 dB) and almost no effect from the ground plane size.
1.4 Mesh #
- Global: 15-20 lines per wavelength.
- Local refinement (maximum step width) around the thin pin and the cable: about 0.3-0.5 mm.
- Check the port: the line impedance in the port mode information must be about 50 Ω. In this model it was 50.17 Ω. If it is off by more than a few tenths of an ohm, the mesh in the cable cross-section is too coarse.
- Check the mesh view with a cut through the pin and the patch. Very thin sheets (0.6 mm) with a pin ending just inside the patch give a poorly resolved junction, so let the pin pass through the patch.
1.5 Tuning with the Smith chart #
Read Z at 869 MHz from the Smith chart (Z, not S11).
- Goal: R = 50 Ω, X = 0 Ω. R between 40 and 60 Ω with |X| below about 10 Ω already gives S11 better than -15 dB.
- R too low: move the probe outward (larger
fy). - X positive (inductive): mainly the probe pin. A larger
fyreduces it. - X negative (capacitive): the patch is too short, so lengthen
pl. - Rough sensitivities for the final geometry: about +8 Ω of X per +1 mm of
pland only about +2 Ω of R. Change only one parameter per run, otherwise the effects can't be separated. - Bandwidth: the -10 dB bandwidth is about 42 MHz, so the 2 MHz of the band is not critical.
Tuning history of the final geometry (all at 869 MHz):
| Step | fy (mm) |
pl (mm) |
Z (Ω) |
|---|---|---|---|
| RG-401 feed, 1.5 mm sheet, radome | 50 | 151 | 51.1 - j1.3 |
| Sheets reduced to 0.6 mm | 50 | 151 | 47.1 - j12.1 |
| Pin ends flush with the patch top | 50 | 151 | 48.2 - j8.0 |
fy and pl changed |
47 | 150.5 | 52.3 - j9.0 |
pl only |
47 | 151.5 | 54.3 - j0.9 |
| Real pin (stick-out and fillet) | 47 | 151.5 | 53.2 - j1.9 |
1.6 Simulation results #
| Quantity | Value |
|---|---|
| Input impedance at 869 MHz | 54.3 - j0.9 Ω |
| S11 at 869 MHz | about -26 dB |
| -10 dB bandwidth | 42.2 MHz |
| Gain (theta = 0°) | 9.61 dBi |
| 3 dB width, H-plane (Phi = 0°) | 67.8° |
| 3 dB width, E-plane (Phi = 90°) | 54° |
| Gain at theta = 180° | -12.3 dBi (front-to-back about 22 dB) |
- Metal: with lossy zinc or bare steel instead of PEC the gain changes by about 0.01 dB. Galvanized sheet is fine.
- Ground plane width:
gwbetween 200 and 350 mm barely changes the match (S11 -20 dB or better). A narrower plane gives a wider H-plane beam (72° at 200 mm) and a lower gain (9.3 dBi at 200 mm, 10.1 dBi at 350 mm in an earlier model without radome). 250 mm is a compromise.








Part 2: The real build #
2.1 Parts and tools #
| Item | Notes |
|---|---|
| Patch sheet | zinc-coated steel, 0.6 mm, 151.5 x 190 mm |
| Ground plane sheet | zinc-coated steel, 0.6 mm, 250 x 350 mm |
| Feed point | SMA Square Flange Jack Receptacle (Exposed Teflon), length ~15mm |
| Standoffs | dielectric (e.g. nylon) for a 12 mm gap, one at each patch corner |
| Radome | PVC, 2 mm wall, held exactly 15 mm above the patch |
| Optional | ferrite or sleeve choke on the cable |
| Paint | Urethan 71 for covering bare metal sheets |
2.2 Step by step #
Step 1: Cut the sheets #
- Cut the patch to 151.5 x 190 mm and the ground plane to 250 x 350 mm. Keep the edges straight and the corners square. Deburr the edges.
- The patch resonance moves by about 6 MHz per mm of the 151.5 mm length, so cut that dimension carefully. The 190 mm side matters much less.
- Touch up with Urethan 71 paint, but keep the contact areas bare.

Step 2: Drill the holes #
- Patch: the probe hole lies on the center line across the 190 mm width (95 mm from each long edge). Along the 151.5 mm length it is 47 mm from the center, which is 28.75 mm from the nearer edge and 122.75 mm from the other. The hole is slightly larger than the SMA flange jack pin (about 1.8 mm). The patch has 3mm holes 10 mm x 10 mm from edges for the standoffs.
- Ground plane: the plane is centered under the patch. Drill the hole for the SMA flange jack (exposed Teflon diameter) so that it lies directly below the probe point, 47 mm from the plane's center along the 350 mm axis.
- Mark the four standoff positions near the patch corners (10 mm x 10 mm from edges).
- The 350 mm side of the ground plane is parallel to the 151.5 mm patch axis (the feed axis).
Step 3: Prepare the SMA flange jack #
- Strip the exposed Teflon/PTFE so that the it reaches the patch underside exactly: the PTFE end must be 12 mm above the top of the ground plane, and the inner conductor sticks out about 1 mm above the patch after the patch is fitted.
- The PTFE length sets the height
h, which is the most sensitive dimension of the whole antenna. - Keep the SMA exposed PTFE end square and clean.

Step 4: Mount the SMA flange jack in the ground plane #
- The SMA flange jack is mounted with four 2mm countersunk screws in the ground plane.
Step 5: Assemble the patch #
- Fit the patch over the pin and onto the PTFE, and fix it on the standoffs at the corners. Check the gap
h= 12 mm with a caliper at several points. A thin sheet may bow, so make sure it sits flat. - Solder the pin to the patch on top. Use a hot iron and a short contact time, so the sheet does not warp and the PTFE does not shrink. Keep the amount of solder small and the same for every unit, because the stick-out and the fillet shift the match by a few ohms.
- Keep the patch centered over the ground plane.


Step 6: Radome #
- Fit the PVC cover so that its inner roof is exactly 15 mm above the patch, with a 2 mm wall. Never let it touch the patch.
- The cover shifts the resonance by several MHz, so wall thickness and spacing must be the same on every copy.


2.3 Measuring and trimming #
- Calibrate the FSH6 at the antenna connector (or use port extension), otherwise the cable rotates the Smith chart.
- Measure S11 without the radome first, then with it, and keep the antenna away from your body and from large metal objects.
- Check the cable: move the cable and put a hand near the feed. If S11 changes noticeably, add a choke.
- Trimming: a longer
pllowers the resonance by about 6 MHz per mm (simulation estimate). Remove material from or add length to the radiating edges (the two 190 mm edges at the ends of the 151.5 mm length) symmetrically, in steps of 0.5 mm, and measure again after every step. - Measure again at the final mounting position.

2.4 Results #
| Simulation (3 mm radome) | Measured with radome (2 mm) | Measured without radome | |
|---|---|---|---|
| S11 at 869 MHz | -29 dB | -28.2 dB | -18.3 dB |
| -10 dB bandwidth | 42.5 MHz | 52 MHz | 49 MHz |
| Minimum | near 869 MHz | -45.7 dB at 872 MHz | -44.3 dB at 878 MHz |




Observations:
- The band 868-870 MHz is far inside the -10 dB bandwidth, so the build needs no trimming.
- The radome moves the dip from 878 MHz down to 872 MHz and improves the match at 869 MHz from -18 dB to -28 dB. In the simulation the wall was 3 mm, in the build it is 2 mm, so a smaller shift would be expected in the model. This is plausible with the typical PVC values (εr about 3), but check it in CST by simulating both cases.
- If you want the dip exactly at 869 MHz, lengthen the patch by about 0.5 mm. It is not needed, and the current offset leaves headroom for rain or ice on the cover, which pulls the resonance down further.
- The real bandwidth is slightly larger than simulated, which means a slightly lower Q (PVC loss, solder joints, connector). Gain and pattern were not measured. A quick plausibility test is a comparison of the received level with a sleeve dipole at a fixed distance. In the main beam the patch should be about 7 dB stronger.
2.5 Build tolerances and tips #
- Most sensitive: the height
h(PTFE length and flat patch), thenpl. The probe positionfyis not critical to about ±1 mm. - Keep radome wall thickness and spacing identical on every copy.
- Mount the antenna with the 151.5 mm axis (and the feed axis) vertical, so the beam is about 68° wide horizontally and 54° vertically.
- Use clean, firm contacts everywhere. The bulk metal does not matter, but the joints (pin, cable tube, any screws) can add loss and corrode.
915 MHz version (simulation only, not built) #
For the US ISM band (902-928 MHz), a scaled version was simulated on request from the Meshtastic community. It has not been built or measured, so treat it as a starting point and expect to retune.


| Parameter | 868/869 MHz (built) | 915 MHz (simulated) |
|---|---|---|
Patch length pl |
151.5 mm | 144 mm |
Patch width pw |
190 mm | 180 mm |
Height h |
12 mm | 11 mm |
Probe offset fy |
47 mm | 45 mm |
Ground plane gw x gl |
250 x 350 mm | 235 x 330 mm |
| Radome spacing | 15 mm | 15 mm |
| Z at center frequency | 53.2 - j1.9 Ω | 50.12 + j0.14 Ω |
| S11 at center frequency | about -29 dB | -43.0 dB |
| -10 dB bandwidth | 42.5 MHz | 45.56 MHz |
| Gain (theta = 0°) | 9.61 dBi | 9.63 dBi |
| 3 dB width, H-plane (Phi = 0°) | 67.9° | 67.9° |
| 3 dB width, E-plane (Phi = 90°) | 54° | 54.2° |
| Gain at theta = 180° | -12.3 dBi | -11.9 dBi |
| Front-to-back | about 22 dB | about 21.5 dB |
The 915 MHz design is essentially the 869 MHz antenna scaled by about 0.95, and the pattern and gain come out the same within a fraction of a dB and degree.



Notes for anyone building it:
- The simulation used a 3 mm PVC wall, 15 mm above the patch. In the 869 MHz build, a 2 mm wall moved the dip about 3 MHz higher than simulated. Expect a similar shift at 915 MHz, or model your own wall thickness first and adjust
pl. - The sensitive dimensions are the same as before: the height
h(the PTFE of the SMA flange jack must end exactly 11 mm above the ground plane) and the patch lengthpl. - Drilling position of the probe: 45 mm from the patch center, which is 27 mm from the nearer edge (72 - 45) and 117 mm from the other.
- Check your local rules for antenna gain and transmit power. With about 9.6 dBi, the conducted power may have to be reduced in some regions.
👈 Home
